U.S. patent number 4,695,964 [Application Number 06/676,921] was granted by the patent office on 1987-09-22 for image correction processing method.
This patent grant is currently assigned to Hitachi, Ltd.. Invention is credited to Koichi Homma, Fuminobu Komura, Hideo Ota, Youichi Seto.
United States Patent |
4,695,964 |
Seto , et al. |
September 22, 1987 |
Image correction processing method
Abstract
Distortion models consisting of two forward and backward
reciprocating planes that let a corrected image correspond to an
uncorrected image for each scanning direction are constituted to
let an uncorrected satellite image scanned by a reciprocating
scanning sensor with scanning overlap or scanning underlap
correspond to a corrected image after distortion correction, by
means of a continuous mapping function on a 1:1 basis. It is
determined whether each point on the corrected image is a point on
an overlapping scan or on an underlapping scan or on a normal scan
from the existence of a real point on the distortion models.
Inventors: |
Seto; Youichi (Hadano,
JP), Homma; Koichi (Sagamihara, JP),
Komura; Fuminobu (Yokohama, JP), Ota; Hideo
(Hitachi, JP) |
Assignee: |
Hitachi, Ltd. (Tokyo,
JP)
|
Family
ID: |
16851653 |
Appl.
No.: |
06/676,921 |
Filed: |
November 30, 1984 |
Foreign Application Priority Data
|
|
|
|
|
Dec 2, 1983 [JP] |
|
|
58-226857 |
|
Current U.S.
Class: |
345/427;
348/147 |
Current CPC
Class: |
G01C
11/025 (20130101) |
Current International
Class: |
G01C
11/02 (20060101); G01C 11/00 (20060101); G01C
007/02 (); G03B 027/68 (); G06F 015/31 () |
Field of
Search: |
;364/514,518,571
;343/5CM ;358/109 ;382/42,54 ;355/52 ;356/1,2 |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
Tsuchia et al.: High Precision All Digital LANDSAT Image Data
Processing Techniques, Hitachi Review, vol. 29, No. 5, 1980, pp.
267-272. .
Ihara et al.: High Precision Modeling of LANDSAT Imagery
Distortion, ISTC, Jul. 1982, pp. 1-18..
|
Primary Examiner: Gruber; Felix D.
Attorney, Agent or Firm: Antonelli, Terry & Wands
Claims
What is claimed is:
1. A satellite image geometric correction processing method for use
in an image correction system for correcting distortion in an image
produced by displaying received data resulting from reciprocating
scanning of an image detecting device mounted on a satellite,
comprising the steps of:
establishing a first coordinate system for a corrected image;
allocating data signals representing received image data to second
and third coordinate systems to provide uncorrected images for the
forward and backward scanning directions of said image detecting
device, respectively;
locating the respective coordinate points of the received image
data in said second and third coordinate systems which correspond
to the points of said first coordinate system;
determining for each point of said first coordinate system whether
or not the corresponding points of said second and third coordinate
systems exist on a scanning region in the respective coordinate
system;
determining the intensity of image data at a point of said first
coordinate system through interpolation by assuming that the point
exists on an underlapping scanning region when the corresponding
coordinate point of said second and third coordinate system does
not exist on a scanning region of either of said second and third
coordinate system; and
visually producing the corrected image data in said first
coordinate system on the basis of the determined intensity of each
point therein.
2. A satellite image geometric correction processing method
according to claim 1, which further comprises a step of determining
the intensity of image data at a point of said first coordinate
system through interpolation by assuming that the point exists on
an overlapping scanning region when the corresponding coordinate
points of said second and third coordinate systems exist on a
scanning region of both of these coordinate systems.
3. A satellite image geometric correction processing method
according to claim 2, wherein said second coordinate system is
formed by alternate forward scanning data regions and imaginary
forward scanning data regions and said third coordinate system is
formed by alternate backward scanning data regions and imaginary
backward scanning data regions.
4. A satellite image geometric correction processing method
according to claim 1, wherein said second coordinate system is
formed by alternate forward scanning data regions and imaginary
forward scanning data regions and said third coordinate system is
formed be alternate backward scanning data regions and imaginary
backward scanning data regions.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to an image correction processing method for
a satellite image having geometric distortion, and more
particularly to a correction processing method for the image
obtained by a sensor TM (Thematic Mapper; optical sensor) mounted
on the Landsat 4 Satellite.
2. Description of the Prior Art
Scanning error resulting from orbit and attitude fluctuation in the
conventional satellite image is within the allowable error inside a
model, and correcting scanning error is not taken into
consideration because scanning is unidirectional scanning as
typified by MSS (multispectral scanner) and resolution is as low as
2.7.times. when compared with a TM sensor.
Therefore, the prior art technique involves a problem in that in a
model (mapping function) which gives correspondence between
corrected image coordinates and uncorrected image coordinates; the
mapping function is discontinued at the positions where a scanning
error exists; hence, the image cannot be expressed at said error
positions.
SUMMARY OF THE INVENTION
The present invention is directed to providing a correction
processing method which can easily distinguish scanning distortion
caused by overlap or underlap between points on the satellite image
read by a reciprocating scanning sensor.
To accomplish the object described above, the present invention
constitutes two uncorrected image coordinate systems for forward
and backward scanning directions, respectively, in which the
corrected image coordinates correspond to the uncorrected image
coordinates for the scanning directions, and which carries out
processing in such a manner that if no real point is found on
either of the two uncorrected image coordinate systems for a point
on the corrected image, the point is judged to be on an
underlapping scan; or, if a real point is found on both uncorrected
image coordinate systems, the point is judged to be on an
overlapping scan; and finally, if a real point is found on only one
of the two uncorrected image coordinate systems, the point is
judged to be on a normal scan.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 shows a conventional system representing a corrected image
coordinate system and an uncorrected image coordinate system;
FIG. 2 shows the Landsat 4 Satellite schematically;
FIG. 3 shows coordinate systems of the present invention
representing a corrected image coordinate system and two
uncorrected image coordinate systems; and
FIG. 4 shows a flow chart for correction processing of the Landsat
4 Satellite TM image in accordance with the corresponding two-plane
distortion model system.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
First of all, we will consider the reason why conventional
distortion correction systems can not judge and correct scanning
error (overlapping scan 1 and underlapping scan 2 shown in FIG. 1)
distortion with reference to FIGS. 1 and 2. FIG. 2 shows the
outline of the Landsat 4 Satellite TM sensor. An oscillating
scanning mirror 8 reciprocatingly scans the ground surface in the
direction corresponding to that indicated by an arrow 9 to take
photographs of the ground surface. The image obtained by
photographing the ground surface exhibits differences between
forward scanning 6 and backward scanning 7 due to the attitude
fluctuation of the scanning mirror and satellite as represented by
the corrected image coordinate system 4 in FIG. 1. The following
processes (1) through (3) are carried out in order to correct the
received image.
(1) The mapping function .phi. representing geometric
correspondence from the received uncorrected image coordinate
system 3 to the corrected image coordinate system 4 is determined.
The mapping function .phi. is determined from data such as the
orbit attitude data of the satellite, the scanning angle, and the
like.
(2) A representative point on the corrected image coordinate system
4 such as a point (x.sub.i, y.sub.i) on the received image
corresponding to the normal grid point (u.sub.i, v.sub.i), for
example, is obtained by repeated calculation of convergence of the
mapping function .phi. (primarily because the inverse mapping
function .phi..sup.-1 can not be determined), and the points
corresponding to those other than the representative point are
interpolated to approximate the inverse mapping function
.phi..sup.-1.
(3) The point of the corrected pixel position (u, v) is obtained
from the corresponding point (x, y) with the approximate inverse
mapping function .phi..sup.-1, and the surrounding received image
data is interpolated to obtain a corrected image intensity value,
since its position does not generally correspond to the pixel
position on the received image.
If the mapping function .phi. is continuous, the distortion
correction processing described above can approximate the inverse
mapping function .phi..sup.-1 with an arbitrary level of accuracy
by increasing the density of the representative point on the output
image; hence, it does not present any problem. When the received
image coordinate system 3 has an overlapping scan 1 or an
underlapping scan 2, however, the mapping function .phi. is a
many-to-one or a zero-to-one relation and approximation with the
continuous function is impossible.
Hereinafter, one embodiment of the present invention will be
described with reference to the reciprocating two-plane distortion
model of the Landsat 4 Satellite by referring to FIGS. 3 and 4.
The mapping function .phi. representing the correspondence between
the corrected image coordinate system 4 and the uncorrected image
coordinate system 3 is the function that uses the attitude angle
.theta.(t) of the satellite, the position .gamma.(t) and the
scanning angle .beta.(t) as its variables. Mapping .phi.(x, y) from
the uncorrected image coordinates x-y to the corrected image
coordinates u-v can be expressed by the following formula. Here, t
represents the time, and is a function of the uncorrected image
coordinates x-y as t=t(x, y):
The present invention is practiced with the following [A] and
[B].
[A]: Introduction of double mapping functions .phi..sub.1.sup.-1
and .phi..sub.2.sup.-1
Mapping .phi..sup.-1 from the corrected image coordinate system u-v
to the uncorrected image coordinate system x-y is not 1:1 mapping
on the TM image for the following reasons (1) and (2).
(1) There is a region .OMEGA..sub.1, scanning overlap region, on
the corrected image where mapping is a one-to-many relation.
(2) There is a region .OMEGA..sub.2, the scanning underlap region,
on the corrected image where no corresponding point exists on the
uncorrected image coordinate system x-y.
Therefore, the present invention considers the two coordinates
systems x.sub.1 -y.sub.1 and x.sub.2 -y.sub.2 to be the uncorrected
image coordinate system 3.
These two coordinates are the following (a) and (b) are as shown in
FIG. 3:
(a) The coordinate system x.sub.1 -y.sub.1 formed by alternately
coupling forward scanning data regions 6 and imaginary forward
scanning data regions 16; and
(b) The coordinate system x.sub.2 -y.sub.2 formed by alternately
coupling backward scanning data regions 7 and imaginary backward
scanning data regions 17.
The two coordinate systems define two mappings .phi..sub.1.sup.-1
and .phi..sub.2.sup.-1 corresponding to x.sub.1 -y.sub.1 and
x.sub.2 -y.sub.2, respectively. When carrying out imaginary forward
(backward) scanning, mapping is obtained by proceeding as if
scanning were made in practice forward (or backward) scanning with
forward (or backward) scanning characteristics. Therefore, mappings
.phi..sub.1.sup.-1 and .phi..sub.2.sup.-1 are continuous, 1:1
mapping functions.
The following can be judged from the relation between four kinds of
points a, b, c, d on the corrected image coordinate system 4 and
the corresponding points a.sub.1, b.sub.1, c.sub.1, d.sub.1,
a.sub.2, b.sub.2, c.sub.2, d.sub.2 on the uncorrected image:
(i) The points a.sub.1, a.sub.2 corresponding to the point a on the
scanning overlap region .OMEGA..sub.1 exist in the practical image
data regions 6, 7 on the uncorrected image coordinate system 3;
and
(ii) The points c.sub.1, c.sub.2 corresponding to the point c on
the scanning underlap region .OMEGA..sub.2 exist in the imaginary
image data regions 16, 17 on the uncorrected image 2.
[B] Introduction of the reciprocating two-plane distortion model
correction system:
FIG. 4 shows the flow of the reciprocating two-plane distortion
model correction processing.
Step 18: The coordinates (u, v) on the corrected image coordinate
system 4 are determined.
Step 19: The corresponding points (x.sub.1, y.sub.1), (x.sub.2,
y.sub.2) on the uncorrected image coordinate system 3 are
determined from the point (u, v) on the corrected image coordinate
system 4 by mapping .phi..sub.1.sup.-1 (u, v) and mapping
.phi..sub.2.sup.-1 (u, v). The satellite parameter data 20 are used
when determining the mapping .phi..sub.1.sup.-1 and
.phi..sub.2.sup.-1.
Step 20: Data such as the satellite position, the attitude, the
scanning angle of the sensor, and the like are calculated.
Step 21: It is determined whether or not the corresponding point
(x.sub.1, y.sub.1) exists on a real scan. If it does step 22 is
followed, and if not, step 23 is followed.
Step 22: It is determined whether or not the corresponding point
(x.sub.2, y.sub.2) exists on the real scan. If it does step 24 is
followed and if not, step 25 is followed.
Step 23: It is determined whether or not corresponding point
(x.sub.2, y.sub.2) exists on the real scan. If so, step 26 is
followed and if not, step 27 is followed.
Step 24: Interpolation is made assuming that the point exists on an
overlapping scan.
Step 25: Interpolation is made assuming that the point exists on a
normal scan of the coordinates expressed by .phi..sub.1.sup.-1.
Step 26: interpolation is made assuming that the point exists on a
normal scan of the coordinates expressed by .phi..sub.2.sup.-1.
Step 27: Interpolation is made assuming that the point exists on an
underlapping scan.
After the procedures described above have been carried out for all
points (u, v) on the corrected image, geometric distortion taking
the scanning error into consideration can be corrected.
The present invention is particularly effective for detecting
scanning error when correcting a satellite image having geometric
distortion such as scanning overlap or underlap resulting from
reciprocating scanning.
* * * * *